Atlas

Compare Models

Gemini 3.8 Flash vs Muse Ember

Muse Ember leads by 1.5 capability points (≈95% interval −0.8 to +3.9): statistically tied · P(higher) ≈ 90%

Wins 36 of 57 shared benchmarks (20 lost, 1 tied)

Capability vs. relative cost
Gemini 3.8 FlashMuse EmberClosed weightsOpen weights

Benchmark wins20 – 36 · 57 shared, 1 tied

Comparing matching setups only (1 benchmark excluded) · suites counted once
← Gemini 3.8 FlashMuse Ember →
Data: Artificial Analysis, Vals AI, Surge AI and 10 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Muse Ember minus Gemini 3.8 Flash.

HiL-Bench (Human-in-Loop Benchmark)Tool & computer use › General agents41.574.0+32.5
DAXBench v2Software › Data & ML87.262.8−24.5
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering19.740.9+21.2
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax4.524.2+19.7
Terminal-Bench 4.0Software › Software engineering19.739.4+19.7
AA-Briefcase v1.1 EloProfessional › General knowledge work12041526+17.7
APEX-SWESoftware › Software engineering36.452.9+16.5
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine2.017.2+15.2
HLAB-Vals-1.1Professional › Legal12.125.0+12.9
Artificial Analysis Coding Agent Index v2.0Software › Software engineering33.345.9+12.7
BuseyBench SVGPerception & media › Generation6.65.4−12.3
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge34.322.0−12.3
Harvey LAB (Mercor run, undisclosed ~213-task cohort)Professional › Legal11.422.7+11.3
FORTRESS Average Risk Score (ARS)Safety & alignment22.711.9+10.8
Legal Research BenchProfessional › Legal38.949.0+10.1
SimpleBenchKnowledge & reasoning › General knowledge82.472.7−9.7
Terminal-Bench-Science 0.1Science › Multidisciplinary10.520.0+9.5
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering38.247.1+8.9
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering18.727.2+8.4
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge23.77.7−8.0
Ramp SWE-Bench v2 (Bash harness)Software › Software engineering47.054.4+7.4
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations60.968.1+7.2
Professional Reasoning Benchmark - LegalProfessional › Legal48.955.6+6.7
Professional Reasoning Benchmark - FinanceProfessional › Finance, accounting & tax49.556.0+6.5
Vals Legal IndexProfessional › Legal69.875.7+5.9
Artificial Analysis Intelligence Index v5.0.1Composite indexes30.836.0+5.2
Vibe Code Bench v1.1Software › Frontend & design78.783.7+5.1
GDP.pdf (November 2026 rescoring)Professional › General knowledge work44.249.2+5.0
Tax Agent BenchProfessional › Finance, accounting & tax66.871.4+4.6
Vals Index v2.2Composite indexes56.060.5+4.5
τ³-Banking (Mercor run, bm25_grep retrieval)Tool & computer use › Customer service & operations48.853.3+4.5
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following63.258.8−4.4
TaxCalcBench TY25Professional › Finance, accounting & tax6.02.0−4.0
LMArena Agent Arena Steering BurdenTool & computer use › General agents-2.84.8+3.8
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations58.562.1+3.6
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents4.0-2.7−3.4
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations26.529.8+3.3
RiemannBench (November 2026 rescoring)Math56.853.6−3.2
LMArena Code Arena WebDevSoftware › Frontend & design15851608+3.2
Artificial Analysis Progress Index v1Composite indexes320329+3.0
GDP.xlsxProfessional › General knowledge work18.315.4−2.9
CritPt v2 (agentic, avg@5)Science › Physical sciences65.162.3−2.9
APEX-Agents 1.1Professional › General knowledge work72.975.7+2.8
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine15.618.3+2.8
Chartography (November 2026 rescoring)Perception & media › Documents & charts81.779.3−2.4
Finance Agent v2.1Professional › Finance, accounting & tax65.863.8−1.9
APEX-1 (Mercor, mean score)Professional › General knowledge work68.870.4+1.6
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents1.0-2.0−1.5
SWE Atlas v1.1 - OverallSoftware › Software engineering57.356.0−1.3
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents8.96.6−1.1
ParseBench-2Perception & media › Documents & charts75.574.5−1.0
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents1.51.0−1.0
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax72.972.2−0.7
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling0.00.9+0.5
Vals Finance IndexProfessional › Finance, accounting & tax60.560.9+0.5
GBENCH One-shot Coding GScoreGames56.656.4−0.3
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work22.822.80.0
Data: Artificial Analysis, Vals AI, Surge AI and 10 othersatlas.village.dev
Win counts use each benchmark’s declared score direction. Represented suite slices and the derived Capability score are excluded from headline totals. Missing or incompatible setups are excluded, never counted as losses.
Loading Atlas data…